409 – Time Series, Longitudinal, and Index Data
Time-Series Cross-Sectional Approach for Small-Area Poverty Models
Wesley Basel
U.S. Census Bureau
Jasen Taciak
U.S. Census Bureau
Current production models of poverty utilized by the Census Bureau for the Small Area Income and Poverty Estimates (SAIPE) program generally incorporate only a single-year of inputs. These models produce parameter estimates by assuming some degree of homogeneity across areas. With six years of consistent inputs to the SAIPE model, including the American Community Survey, we find that additional precision is obtained by modifying assumptions of homogeneity in parameter estimates across years, and relaxing some cross-sectional assumptions. The impact of these alternative assumptions is then illustrated in terms of the change in precision to the estimate within population clusters and within the context of the current production SAIPE models.